[PATCH 4.19 0/5] backport some more fscrypt fixes to 4.19

[PATCH 4.19 0/5] backport some more fscrypt fixes to 4.19

[Eric Biggers]

Backport some fscrypt fixes from upstream 5.2 to 4.19-stable.

This is needed to get 'kvm-xfstests -c ext4,f2fs,ubifs -g encrypt' to
fully pass on 4.19-stable.  Before, generic/397 and generic/429 failed
on UBIFS due to missing "fscrypt: fix race where ->lookup() marks
plaintext dentry as ciphertext".

This also fixes some bugs that aren't yet covered by the xfstests.
E.g., "fs, fscrypt: clear DCACHE_ENCRYPTED_NAME when unaliasing
directory" fixes a bug that caused real-world problems on Chrome OS.

Some relatively straightforward adjustments were needed to the patches,
mainly due to the refactoring of fscrypt.h that was done in 5.1.

Eric Biggers (5):
  fscrypt: clean up and improve dentry revalidation
  fscrypt: fix race allowing rename() and link() of ciphertext dentries
  fs, fscrypt: clear DCACHE_ENCRYPTED_NAME when unaliasing directory
  fscrypt: only set dentry_operations on ciphertext dentries
  fscrypt: fix race where ->lookup() marks plaintext dentry as
    ciphertext

 fs/crypto/crypto.c              | 58 +++++++++++++------------
 fs/crypto/fname.c               |  1 +
 fs/crypto/hooks.c               | 28 ++++++++----
 fs/dcache.c                     | 15 +++++++
 fs/ext4/ext4.h                  | 62 ++++++++++++++++++++-------
 fs/ext4/namei.c                 | 76 ++++++++++++++++++++++-----------
 fs/f2fs/namei.c                 | 17 +++++---
 fs/ubifs/dir.c                  |  8 ++--
 include/linux/dcache.h          |  2 +-
 include/linux/fscrypt.h         | 30 +++++++------
 include/linux/fscrypt_notsupp.h |  9 ++--
 include/linux/fscrypt_supp.h    |  6 ++-
 12 files changed, 205 insertions(+), 107 deletions(-)

-- 
2.29.1

[PATCH 4.19 1/5] fscrypt: clean up and improve dentry revalidation

[Eric Biggers]

From: Eric Biggers <ebiggers@google.com>

commit 6cc248684d3d23bbd073ae2fa73d3416c0558909 upstream.

Make various improvements to fscrypt dentry revalidation:

- Don't try to handle the case where the per-directory key is removed,
  as this can't happen without the inode (and dentries) being evicted.

- Flag ciphertext dentries rather than plaintext dentries, since it's
  ciphertext dentries that need the special handling.

- Avoid doing unnecessary work for non-ciphertext dentries.

- When revalidating ciphertext dentries, try to set up the directory's
  i_crypt_info to make sure the key is really still absent, rather than
  invalidating all negative dentries as the previous code did.  An old
  comment suggested we can't do this for locking reasons, but AFAICT
  this comment was outdated and it actually works fine.

Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
---
 fs/crypto/crypto.c      | 58 +++++++++++++++++++++--------------------
 fs/crypto/hooks.c       |  4 +--
 include/linux/dcache.h  |  2 +-
 include/linux/fscrypt.h |  6 ++---
 4 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/fs/crypto/crypto.c b/fs/crypto/crypto.c
index c83ddff3ff4ac..04a3c2c92b21e 100644
--- a/fs/crypto/crypto.c
+++ b/fs/crypto/crypto.c
@@ -314,45 +314,47 @@ int fscrypt_decrypt_page(const struct inode *inode, struct page *page,
 EXPORT_SYMBOL(fscrypt_decrypt_page);
 
 /*
- * Validate dentries for encrypted directories to make sure we aren't
- * potentially caching stale data after a key has been added or
- * removed.
+ * Validate dentries in encrypted directories to make sure we aren't potentially
+ * caching stale dentries after a key has been added.
  */
 static int fscrypt_d_revalidate(struct dentry *dentry, unsigned int flags)
 {
 	struct dentry *dir;
-	int dir_has_key, cached_with_key;
+	int err;
+	int valid;
+
+	/*
+	 * Plaintext names are always valid, since fscrypt doesn't support
+	 * reverting to ciphertext names without evicting the directory's inode
+	 * -- which implies eviction of the dentries in the directory.
+	 */
+	if (!(dentry->d_flags & DCACHE_ENCRYPTED_NAME))
+		return 1;
+
+	/*
+	 * Ciphertext name; valid if the directory's key is still unavailable.
+	 *
+	 * Although fscrypt forbids rename() on ciphertext names, we still must
+	 * use dget_parent() here rather than use ->d_parent directly.  That's
+	 * because a corrupted fs image may contain directory hard links, which
+	 * the VFS handles by moving the directory's dentry tree in the dcache
+	 * each time ->lookup() finds the directory and it already has a dentry
+	 * elsewhere.  Thus ->d_parent can be changing, and we must safely grab
+	 * a reference to some ->d_parent to prevent it from being freed.
+	 */
 
 	if (flags & LOOKUP_RCU)
 		return -ECHILD;
 
 	dir = dget_parent(dentry);
-	if (!IS_ENCRYPTED(d_inode(dir))) {
-		dput(dir);
-		return 0;
-	}
-
-	spin_lock(&dentry->d_lock);
-	cached_with_key = dentry->d_flags & DCACHE_ENCRYPTED_WITH_KEY;
-	spin_unlock(&dentry->d_lock);
-	dir_has_key = (d_inode(dir)->i_crypt_info != NULL);
+	err = fscrypt_get_encryption_info(d_inode(dir));
+	valid = !fscrypt_has_encryption_key(d_inode(dir));
 	dput(dir);
 
-	/*
-	 * If the dentry was cached without the key, and it is a
-	 * negative dentry, it might be a valid name.  We can't check
-	 * if the key has since been made available due to locking
-	 * reasons, so we fail the validation so ext4_lookup() can do
-	 * this check.
-	 *
-	 * We also fail the validation if the dentry was created with
-	 * the key present, but we no longer have the key, or vice versa.
-	 */
-	if ((!cached_with_key && d_is_negative(dentry)) ||
-			(!cached_with_key && dir_has_key) ||
-			(cached_with_key && !dir_has_key))
-		return 0;
-	return 1;
+	if (err < 0)
+		return err;
+
+	return valid;
 }
 
 const struct dentry_operations fscrypt_d_ops = {
diff --git a/fs/crypto/hooks.c b/fs/crypto/hooks.c
index 56debb1fcf5eb..a9492f75bbe13 100644
--- a/fs/crypto/hooks.c
+++ b/fs/crypto/hooks.c
@@ -101,9 +101,9 @@ int __fscrypt_prepare_lookup(struct inode *dir, struct dentry *dentry)
 	if (err)
 		return err;
 
-	if (fscrypt_has_encryption_key(dir)) {
+	if (!fscrypt_has_encryption_key(dir)) {
 		spin_lock(&dentry->d_lock);
-		dentry->d_flags |= DCACHE_ENCRYPTED_WITH_KEY;
+		dentry->d_flags |= DCACHE_ENCRYPTED_NAME;
 		spin_unlock(&dentry->d_lock);
 	}
 
diff --git a/include/linux/dcache.h b/include/linux/dcache.h
index 0880baefd85f8..02b1b40fea5b1 100644
--- a/include/linux/dcache.h
+++ b/include/linux/dcache.h
@@ -210,7 +210,7 @@ struct dentry_operations {
 
 #define DCACHE_MAY_FREE			0x00800000
 #define DCACHE_FALLTHRU			0x01000000 /* Fall through to lower layer */
-#define DCACHE_ENCRYPTED_WITH_KEY	0x02000000 /* dir is encrypted with a valid key */
+#define DCACHE_ENCRYPTED_NAME		0x02000000 /* Encrypted name (dir key was unavailable) */
 #define DCACHE_OP_REAL			0x04000000
 
 #define DCACHE_PAR_LOOKUP		0x10000000 /* being looked up (with parent locked shared) */
diff --git a/include/linux/fscrypt.h b/include/linux/fscrypt.h
index 9fa48180a1f5f..c83e4dc55c062 100644
--- a/include/linux/fscrypt.h
+++ b/include/linux/fscrypt.h
@@ -145,10 +145,8 @@ static inline int fscrypt_prepare_rename(struct inode *old_dir,
  * filenames are presented in encrypted form.  Therefore, we'll try to set up
  * the directory's encryption key, but even without it the lookup can continue.
  *
- * To allow invalidating stale dentries if the directory's encryption key is
- * added later, we also install a custom ->d_revalidate() method and use the
- * DCACHE_ENCRYPTED_WITH_KEY flag to indicate whether a given dentry is a
- * plaintext name (flag set) or a ciphertext name (flag cleared).
+ * This also installs a custom ->d_revalidate() method which will invalidate the
+ * dentry if it was created without the key and the key is later added.
  *
  * Return: 0 on success, -errno if a problem occurred while setting up the
  * encryption key
-- 
2.29.1

[PATCH 4.19 2/5] fscrypt: fix race allowing rename() and link() of ciphertext dentries

[Eric Biggers]

From: Eric Biggers <ebiggers@google.com>

commit 968dd6d0c6d6b6a989c6ddb9e2584a031b83e7b5 upstream.

Close some race conditions where fscrypt allowed rename() and link() on
ciphertext dentries that had been looked up just prior to the key being
concurrently added.  It's better to return -ENOKEY in this case.

This avoids doing the nonsensical thing of encrypting the names a second
time when searching for the actual on-disk dir entries.  It also
guarantees that DCACHE_ENCRYPTED_NAME dentries are never rename()d, so
the dcache won't have support all possible combinations of moving
DCACHE_ENCRYPTED_NAME around during __d_move().

Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
---
 fs/crypto/hooks.c               | 12 +++++++++++-
 include/linux/fscrypt.h         |  2 +-
 include/linux/fscrypt_notsupp.h |  4 ++--
 include/linux/fscrypt_supp.h    |  3 ++-
 4 files changed, 16 insertions(+), 5 deletions(-)

diff --git a/fs/crypto/hooks.c b/fs/crypto/hooks.c
index a9492f75bbe13..2e7498a821a48 100644
--- a/fs/crypto/hooks.c
+++ b/fs/crypto/hooks.c
@@ -49,7 +49,8 @@ int fscrypt_file_open(struct inode *inode, struct file *filp)
 }
 EXPORT_SYMBOL_GPL(fscrypt_file_open);
 
-int __fscrypt_prepare_link(struct inode *inode, struct inode *dir)
+int __fscrypt_prepare_link(struct inode *inode, struct inode *dir,
+			   struct dentry *dentry)
 {
 	int err;
 
@@ -57,6 +58,10 @@ int __fscrypt_prepare_link(struct inode *inode, struct inode *dir)
 	if (err)
 		return err;
 
+	/* ... in case we looked up ciphertext name before key was added */
+	if (dentry->d_flags & DCACHE_ENCRYPTED_NAME)
+		return -ENOKEY;
+
 	if (!fscrypt_has_permitted_context(dir, inode))
 		return -EXDEV;
 
@@ -78,6 +83,11 @@ int __fscrypt_prepare_rename(struct inode *old_dir, struct dentry *old_dentry,
 	if (err)
 		return err;
 
+	/* ... in case we looked up ciphertext name(s) before key was added */
+	if ((old_dentry->d_flags | new_dentry->d_flags) &
+	    DCACHE_ENCRYPTED_NAME)
+		return -ENOKEY;
+
 	if (old_dir != new_dir) {
 		if (IS_ENCRYPTED(new_dir) &&
 		    !fscrypt_has_permitted_context(new_dir,
diff --git a/include/linux/fscrypt.h b/include/linux/fscrypt.h
index c83e4dc55c062..af52a240c7399 100644
--- a/include/linux/fscrypt.h
+++ b/include/linux/fscrypt.h
@@ -97,7 +97,7 @@ static inline int fscrypt_prepare_link(struct dentry *old_dentry,
 				       struct dentry *dentry)
 {
 	if (IS_ENCRYPTED(dir))
-		return __fscrypt_prepare_link(d_inode(old_dentry), dir);
+		return __fscrypt_prepare_link(d_inode(old_dentry), dir, dentry);
 	return 0;
 }
 
diff --git a/include/linux/fscrypt_notsupp.h b/include/linux/fscrypt_notsupp.h
index ee8b43e4c15a6..15bbbd7c52b8b 100644
--- a/include/linux/fscrypt_notsupp.h
+++ b/include/linux/fscrypt_notsupp.h
@@ -183,8 +183,8 @@ static inline int fscrypt_file_open(struct inode *inode, struct file *filp)
 	return 0;
 }
 
-static inline int __fscrypt_prepare_link(struct inode *inode,
-					 struct inode *dir)
+static inline int __fscrypt_prepare_link(struct inode *inode, struct inode *dir,
+					 struct dentry *dentry)
 {
 	return -EOPNOTSUPP;
 }
diff --git a/include/linux/fscrypt_supp.h b/include/linux/fscrypt_supp.h
index 6456c6b2005f4..b8443a2e47dc4 100644
--- a/include/linux/fscrypt_supp.h
+++ b/include/linux/fscrypt_supp.h
@@ -184,7 +184,8 @@ extern int fscrypt_zeroout_range(const struct inode *, pgoff_t, sector_t,
 
 /* hooks.c */
 extern int fscrypt_file_open(struct inode *inode, struct file *filp);
-extern int __fscrypt_prepare_link(struct inode *inode, struct inode *dir);
+extern int __fscrypt_prepare_link(struct inode *inode, struct inode *dir,
+				  struct dentry *dentry);
 extern int __fscrypt_prepare_rename(struct inode *old_dir,
 				    struct dentry *old_dentry,
 				    struct inode *new_dir,
-- 
2.29.1

[PATCH 4.19 3/5] fs, fscrypt: clear DCACHE_ENCRYPTED_NAME when unaliasing directory

[Eric Biggers]

From: Eric Biggers <ebiggers@google.com>

commit 0bf3d5c1604ecbbd4e49e9f5b3c79152b87adb0d upstream.

Make __d_move() clear DCACHE_ENCRYPTED_NAME on the source dentry.  This
is needed for when d_splice_alias() moves a directory's encrypted alias
to its decrypted alias as a result of the encryption key being added.

Otherwise, the decrypted alias will incorrectly be invalidated on the
next lookup, causing problems such as unmounting a mount the user just
mount()ed there.

Note that we don't have to support arbitrary moves of this flag because
fscrypt doesn't allow dentries with DCACHE_ENCRYPTED_NAME to be the
source or target of a rename().

Fixes: 28b4c263961c ("ext4 crypto: revalidate dentry after adding or removing the key")
Reported-by: Sarthak Kukreti <sarthakkukreti@chromium.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
---
 fs/dcache.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/fs/dcache.c b/fs/dcache.c
index 20370a0997bf9..1897833a46685 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -2712,6 +2712,20 @@ static void copy_name(struct dentry *dentry, struct dentry *target)
 		call_rcu(&old_name->u.head, __d_free_external_name);
 }
 
+/*
+ * When d_splice_alias() moves a directory's encrypted alias to its decrypted
+ * alias as a result of the encryption key being added, DCACHE_ENCRYPTED_NAME
+ * must be cleared.  Note that we don't have to support arbitrary moves of this
+ * flag because fscrypt doesn't allow encrypted aliases to be the source or
+ * target of a rename().
+ */
+static inline void fscrypt_handle_d_move(struct dentry *dentry)
+{
+#if IS_ENABLED(CONFIG_FS_ENCRYPTION)
+	dentry->d_flags &= ~DCACHE_ENCRYPTED_NAME;
+#endif
+}
+
 /*
  * __d_move - move a dentry
  * @dentry: entry to move
@@ -2787,6 +2801,7 @@ static void __d_move(struct dentry *dentry, struct dentry *target,
 	list_move(&dentry->d_child, &dentry->d_parent->d_subdirs);
 	__d_rehash(dentry);
 	fsnotify_update_flags(dentry);
+	fscrypt_handle_d_move(dentry);
 
 	write_seqcount_end(&target->d_seq);
 	write_seqcount_end(&dentry->d_seq);
-- 
2.29.1

[PATCH 4.19 4/5] fscrypt: only set dentry_operations on ciphertext dentries

[Eric Biggers]

From: Eric Biggers <ebiggers@google.com>

commit d456a33f041af4b54f3ce495a86d00c246165032 upstream.

Plaintext dentries are always valid, so only set fscrypt_d_ops on
ciphertext dentries.

Besides marginally improved performance, this allows overlayfs to use an
fscrypt-encrypted upperdir, provided that all the following are true:

    (1) The fscrypt encryption key is placed in the keyring before
	mounting overlayfs, and remains while the overlayfs is mounted.

    (2) The overlayfs workdir uses the same encryption policy.

    (3) No dentries for the ciphertext names of subdirectories have been
	created in the upperdir or workdir yet.  (Since otherwise
	d_splice_alias() will reuse the old dentry with ->d_op set.)

One potential use case is using an ephemeral encryption key to encrypt
all files created or changed by a container, so that they can be
securely erased ("crypto-shredded") after the container stops.

Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
---
 fs/crypto/hooks.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/fs/crypto/hooks.c b/fs/crypto/hooks.c
index 2e7498a821a48..9d8910e86ee5d 100644
--- a/fs/crypto/hooks.c
+++ b/fs/crypto/hooks.c
@@ -115,9 +115,8 @@ int __fscrypt_prepare_lookup(struct inode *dir, struct dentry *dentry)
 		spin_lock(&dentry->d_lock);
 		dentry->d_flags |= DCACHE_ENCRYPTED_NAME;
 		spin_unlock(&dentry->d_lock);
+		d_set_d_op(dentry, &fscrypt_d_ops);
 	}
-
-	d_set_d_op(dentry, &fscrypt_d_ops);
 	return 0;
 }
 EXPORT_SYMBOL_GPL(__fscrypt_prepare_lookup);
-- 
2.29.1

[PATCH 4.19 5/5] fscrypt: fix race where ->lookup() marks plaintext dentry as ciphertext

[Eric Biggers]

From: Eric Biggers <ebiggers@google.com>

commit b01531db6cec2aa330dbc91bfbfaaef4a0d387a4 upstream.

->lookup() in an encrypted directory begins as follows:

1. fscrypt_prepare_lookup():
    a. Try to load the directory's encryption key.
    b. If the key is unavailable, mark the dentry as a ciphertext name
       via d_flags.
2. fscrypt_setup_filename():
    a. Try to load the directory's encryption key.
    b. If the key is available, encrypt the name (treated as a plaintext
       name) to get the on-disk name.  Otherwise decode the name
       (treated as a ciphertext name) to get the on-disk name.

But if the key is concurrently added, it may be found at (2a) but not at
(1a).  In this case, the dentry will be wrongly marked as a ciphertext
name even though it was actually treated as plaintext.

This will cause the dentry to be wrongly invalidated on the next lookup,
potentially causing problems.  For example, if the racy ->lookup() was
part of sys_mount(), then the new mount will be detached when anything
tries to access it.  This is despite the mountpoint having a plaintext
path, which should remain valid now that the key was added.

Of course, this is only possible if there's a userspace race.  Still,
the additional kernel-side race is confusing and unexpected.

Close the kernel-side race by changing fscrypt_prepare_lookup() to also
set the on-disk filename (step 2b), consistent with the d_flags update.

Fixes: 28b4c263961c ("ext4 crypto: revalidate dentry after adding or removing the key")
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
---
 fs/crypto/fname.c               |  1 +
 fs/crypto/hooks.c               | 11 ++---
 fs/ext4/ext4.h                  | 62 ++++++++++++++++++++-------
 fs/ext4/namei.c                 | 76 ++++++++++++++++++++++-----------
 fs/f2fs/namei.c                 | 17 +++++---
 fs/ubifs/dir.c                  |  8 ++--
 include/linux/fscrypt.h         | 22 +++++++---
 include/linux/fscrypt_notsupp.h |  5 ++-
 include/linux/fscrypt_supp.h    |  3 +-
 9 files changed, 139 insertions(+), 66 deletions(-)

diff --git a/fs/crypto/fname.c b/fs/crypto/fname.c
index d7a0f682ca122..17bb9a3fc0b01 100644
--- a/fs/crypto/fname.c
+++ b/fs/crypto/fname.c
@@ -354,6 +354,7 @@ int fscrypt_setup_filename(struct inode *dir, const struct qstr *iname,
 	}
 	if (!lookup)
 		return -ENOKEY;
+	fname->is_ciphertext_name = true;
 
 	/*
 	 * We don't have the key and we are doing a lookup; decode the
diff --git a/fs/crypto/hooks.c b/fs/crypto/hooks.c
index 9d8910e86ee5d..042d5b44f4ed9 100644
--- a/fs/crypto/hooks.c
+++ b/fs/crypto/hooks.c
@@ -104,20 +104,21 @@ int __fscrypt_prepare_rename(struct inode *old_dir, struct dentry *old_dentry,
 }
 EXPORT_SYMBOL_GPL(__fscrypt_prepare_rename);
 
-int __fscrypt_prepare_lookup(struct inode *dir, struct dentry *dentry)
+int __fscrypt_prepare_lookup(struct inode *dir, struct dentry *dentry,
+			     struct fscrypt_name *fname)
 {
-	int err = fscrypt_get_encryption_info(dir);
+	int err = fscrypt_setup_filename(dir, &dentry->d_name, 1, fname);
 
-	if (err)
+	if (err && err != -ENOENT)
 		return err;
 
-	if (!fscrypt_has_encryption_key(dir)) {
+	if (fname->is_ciphertext_name) {
 		spin_lock(&dentry->d_lock);
 		dentry->d_flags |= DCACHE_ENCRYPTED_NAME;
 		spin_unlock(&dentry->d_lock);
 		d_set_d_op(dentry, &fscrypt_d_ops);
 	}
-	return 0;
+	return err;
 }
 EXPORT_SYMBOL_GPL(__fscrypt_prepare_lookup);
 
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 0a4461ac4225f..19109c04710ee 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -2326,23 +2326,47 @@ static inline bool ext4_encrypted_inode(struct inode *inode)
 }
 
 #ifdef CONFIG_EXT4_FS_ENCRYPTION
+static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
+						const struct fscrypt_name *src)
+{
+	memset(dst, 0, sizeof(*dst));
+
+	dst->usr_fname = src->usr_fname;
+	dst->disk_name = src->disk_name;
+	dst->hinfo.hash = src->hash;
+	dst->hinfo.minor_hash = src->minor_hash;
+	dst->crypto_buf = src->crypto_buf;
+}
+
 static inline int ext4_fname_setup_filename(struct inode *dir,
-			const struct qstr *iname,
-			int lookup, struct ext4_filename *fname)
+					    const struct qstr *iname,
+					    int lookup,
+					    struct ext4_filename *fname)
 {
 	struct fscrypt_name name;
 	int err;
 
-	memset(fname, 0, sizeof(struct ext4_filename));
-
 	err = fscrypt_setup_filename(dir, iname, lookup, &name);
+	if (err)
+		return err;
 
-	fname->usr_fname = name.usr_fname;
-	fname->disk_name = name.disk_name;
-	fname->hinfo.hash = name.hash;
-	fname->hinfo.minor_hash = name.minor_hash;
-	fname->crypto_buf = name.crypto_buf;
-	return err;
+	ext4_fname_from_fscrypt_name(fname, &name);
+	return 0;
+}
+
+static inline int ext4_fname_prepare_lookup(struct inode *dir,
+					    struct dentry *dentry,
+					    struct ext4_filename *fname)
+{
+	struct fscrypt_name name;
+	int err;
+
+	err = fscrypt_prepare_lookup(dir, dentry, &name);
+	if (err)
+		return err;
+
+	ext4_fname_from_fscrypt_name(fname, &name);
+	return 0;
 }
 
 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
@@ -2356,19 +2380,27 @@ static inline void ext4_fname_free_filename(struct ext4_filename *fname)
 	fname->usr_fname = NULL;
 	fname->disk_name.name = NULL;
 }
-#else
+#else /* !CONFIG_EXT4_FS_ENCRYPTION */
 static inline int ext4_fname_setup_filename(struct inode *dir,
-		const struct qstr *iname,
-		int lookup, struct ext4_filename *fname)
+					    const struct qstr *iname,
+					    int lookup,
+					    struct ext4_filename *fname)
 {
 	fname->usr_fname = iname;
 	fname->disk_name.name = (unsigned char *) iname->name;
 	fname->disk_name.len = iname->len;
 	return 0;
 }
-static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
 
-#endif
+static inline int ext4_fname_prepare_lookup(struct inode *dir,
+					    struct dentry *dentry,
+					    struct ext4_filename *fname)
+{
+	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
+}
+
+static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
+#endif /* !CONFIG_EXT4_FS_ENCRYPTION */
 
 /* dir.c */
 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c
index 186a2dd05bd87..4191552880bd7 100644
--- a/fs/ext4/namei.c
+++ b/fs/ext4/namei.c
@@ -1343,7 +1343,7 @@ static int is_dx_internal_node(struct inode *dir, ext4_lblk_t block,
 }
 
 /*
- *	ext4_find_entry()
+ *	__ext4_find_entry()
  *
  * finds an entry in the specified directory with the wanted name. It
  * returns the cache buffer in which the entry was found, and the entry
@@ -1353,39 +1353,32 @@ static int is_dx_internal_node(struct inode *dir, ext4_lblk_t block,
  * The returned buffer_head has ->b_count elevated.  The caller is expected
  * to brelse() it when appropriate.
  */
-static struct buffer_head * ext4_find_entry (struct inode *dir,
-					const struct qstr *d_name,
-					struct ext4_dir_entry_2 **res_dir,
-					int *inlined)
+static struct buffer_head *__ext4_find_entry(struct inode *dir,
+					     struct ext4_filename *fname,
+					     struct ext4_dir_entry_2 **res_dir,
+					     int *inlined)
 {
 	struct super_block *sb;
 	struct buffer_head *bh_use[NAMEI_RA_SIZE];
 	struct buffer_head *bh, *ret = NULL;
 	ext4_lblk_t start, block;
-	const u8 *name = d_name->name;
+	const u8 *name = fname->usr_fname->name;
 	size_t ra_max = 0;	/* Number of bh's in the readahead
 				   buffer, bh_use[] */
 	size_t ra_ptr = 0;	/* Current index into readahead
 				   buffer */
 	ext4_lblk_t  nblocks;
 	int i, namelen, retval;
-	struct ext4_filename fname;
 
 	*res_dir = NULL;
 	sb = dir->i_sb;
-	namelen = d_name->len;
+	namelen = fname->usr_fname->len;
 	if (namelen > EXT4_NAME_LEN)
 		return NULL;
 
-	retval = ext4_fname_setup_filename(dir, d_name, 1, &fname);
-	if (retval == -ENOENT)
-		return NULL;
-	if (retval)
-		return ERR_PTR(retval);
-
 	if (ext4_has_inline_data(dir)) {
 		int has_inline_data = 1;
-		ret = ext4_find_inline_entry(dir, &fname, res_dir,
+		ret = ext4_find_inline_entry(dir, fname, res_dir,
 					     &has_inline_data);
 		if (has_inline_data) {
 			if (inlined)
@@ -1405,7 +1398,7 @@ static struct buffer_head * ext4_find_entry (struct inode *dir,
 		goto restart;
 	}
 	if (is_dx(dir)) {
-		ret = ext4_dx_find_entry(dir, &fname, res_dir);
+		ret = ext4_dx_find_entry(dir, fname, res_dir);
 		/*
 		 * On success, or if the error was file not found,
 		 * return.  Otherwise, fall back to doing a search the
@@ -1470,7 +1463,7 @@ static struct buffer_head * ext4_find_entry (struct inode *dir,
 			goto cleanup_and_exit;
 		}
 		set_buffer_verified(bh);
-		i = search_dirblock(bh, dir, &fname,
+		i = search_dirblock(bh, dir, fname,
 			    block << EXT4_BLOCK_SIZE_BITS(sb), res_dir);
 		if (i == 1) {
 			EXT4_I(dir)->i_dir_start_lookup = block;
@@ -1501,10 +1494,50 @@ static struct buffer_head * ext4_find_entry (struct inode *dir,
 	/* Clean up the read-ahead blocks */
 	for (; ra_ptr < ra_max; ra_ptr++)
 		brelse(bh_use[ra_ptr]);
-	ext4_fname_free_filename(&fname);
 	return ret;
 }
 
+static struct buffer_head *ext4_find_entry(struct inode *dir,
+					   const struct qstr *d_name,
+					   struct ext4_dir_entry_2 **res_dir,
+					   int *inlined)
+{
+	int err;
+	struct ext4_filename fname;
+	struct buffer_head *bh;
+
+	err = ext4_fname_setup_filename(dir, d_name, 1, &fname);
+	if (err == -ENOENT)
+		return NULL;
+	if (err)
+		return ERR_PTR(err);
+
+	bh = __ext4_find_entry(dir, &fname, res_dir, inlined);
+
+	ext4_fname_free_filename(&fname);
+	return bh;
+}
+
+static struct buffer_head *ext4_lookup_entry(struct inode *dir,
+					     struct dentry *dentry,
+					     struct ext4_dir_entry_2 **res_dir)
+{
+	int err;
+	struct ext4_filename fname;
+	struct buffer_head *bh;
+
+	err = ext4_fname_prepare_lookup(dir, dentry, &fname);
+	if (err == -ENOENT)
+		return NULL;
+	if (err)
+		return ERR_PTR(err);
+
+	bh = __ext4_find_entry(dir, &fname, res_dir, NULL);
+
+	ext4_fname_free_filename(&fname);
+	return bh;
+}
+
 static struct buffer_head * ext4_dx_find_entry(struct inode *dir,
 			struct ext4_filename *fname,
 			struct ext4_dir_entry_2 **res_dir)
@@ -1563,16 +1596,11 @@ static struct dentry *ext4_lookup(struct inode *dir, struct dentry *dentry, unsi
 	struct inode *inode;
 	struct ext4_dir_entry_2 *de;
 	struct buffer_head *bh;
-	int err;
-
-	err = fscrypt_prepare_lookup(dir, dentry, flags);
-	if (err)
-		return ERR_PTR(err);
 
 	if (dentry->d_name.len > EXT4_NAME_LEN)
 		return ERR_PTR(-ENAMETOOLONG);
 
-	bh = ext4_find_entry(dir, &dentry->d_name, &de, NULL);
+	bh = ext4_lookup_entry(dir, dentry, &de);
 	if (IS_ERR(bh))
 		return (struct dentry *) bh;
 	inode = NULL;
diff --git a/fs/f2fs/namei.c b/fs/f2fs/namei.c
index 4f0cc0c79d1ee..8617e742d0878 100644
--- a/fs/f2fs/namei.c
+++ b/fs/f2fs/namei.c
@@ -432,19 +432,23 @@ static struct dentry *f2fs_lookup(struct inode *dir, struct dentry *dentry,
 	nid_t ino = -1;
 	int err = 0;
 	unsigned int root_ino = F2FS_ROOT_INO(F2FS_I_SB(dir));
+	struct fscrypt_name fname;
 
 	trace_f2fs_lookup_start(dir, dentry, flags);
 
-	err = fscrypt_prepare_lookup(dir, dentry, flags);
-	if (err)
-		goto out;
-
 	if (dentry->d_name.len > F2FS_NAME_LEN) {
 		err = -ENAMETOOLONG;
 		goto out;
 	}
 
-	de = f2fs_find_entry(dir, &dentry->d_name, &page);
+	err = fscrypt_prepare_lookup(dir, dentry, &fname);
+	if (err == -ENOENT)
+		goto out_splice;
+	if (err)
+		goto out;
+	de = __f2fs_find_entry(dir, &fname, &page);
+	fscrypt_free_filename(&fname);
+
 	if (!de) {
 		if (IS_ERR(page)) {
 			err = PTR_ERR(page);
@@ -484,8 +488,7 @@ static struct dentry *f2fs_lookup(struct inode *dir, struct dentry *dentry,
 	}
 out_splice:
 	new = d_splice_alias(inode, dentry);
-	if (IS_ERR(new))
-		err = PTR_ERR(new);
+	err = PTR_ERR_OR_ZERO(new);
 	trace_f2fs_lookup_end(dir, dentry, ino, err);
 	return new;
 out_iput:
diff --git a/fs/ubifs/dir.c b/fs/ubifs/dir.c
index d7c0aa0626cd7..10aab5dccaee4 100644
--- a/fs/ubifs/dir.c
+++ b/fs/ubifs/dir.c
@@ -220,11 +220,9 @@ static struct dentry *ubifs_lookup(struct inode *dir, struct dentry *dentry,
 
 	dbg_gen("'%pd' in dir ino %lu", dentry, dir->i_ino);
 
-	err = fscrypt_prepare_lookup(dir, dentry, flags);
-	if (err)
-		return ERR_PTR(err);
-
-	err = fscrypt_setup_filename(dir, &dentry->d_name, 1, &nm);
+	err = fscrypt_prepare_lookup(dir, dentry, &nm);
+	if (err == -ENOENT)
+		return d_splice_alias(NULL, dentry);
 	if (err)
 		return ERR_PTR(err);
 
diff --git a/include/linux/fscrypt.h b/include/linux/fscrypt.h
index af52a240c7399..c1e4a615bd1cb 100644
--- a/include/linux/fscrypt.h
+++ b/include/linux/fscrypt.h
@@ -32,6 +32,7 @@ struct fscrypt_name {
 	u32 hash;
 	u32 minor_hash;
 	struct fscrypt_str crypto_buf;
+	bool is_ciphertext_name;
 };
 
 #define FSTR_INIT(n, l)		{ .name = n, .len = l }
@@ -138,25 +139,32 @@ static inline int fscrypt_prepare_rename(struct inode *old_dir,
  * fscrypt_prepare_lookup - prepare to lookup a name in a possibly-encrypted directory
  * @dir: directory being searched
  * @dentry: filename being looked up
- * @flags: lookup flags
+ * @fname: (output) the name to use to search the on-disk directory
  *
- * Prepare for ->lookup() in a directory which may be encrypted.  Lookups can be
- * done with or without the directory's encryption key; without the key,
+ * Prepare for ->lookup() in a directory which may be encrypted by determining
+ * the name that will actually be used to search the directory on-disk.  Lookups
+ * can be done with or without the directory's encryption key; without the key,
  * filenames are presented in encrypted form.  Therefore, we'll try to set up
  * the directory's encryption key, but even without it the lookup can continue.
  *
  * This also installs a custom ->d_revalidate() method which will invalidate the
  * dentry if it was created without the key and the key is later added.
  *
- * Return: 0 on success, -errno if a problem occurred while setting up the
- * encryption key
+ * Return: 0 on success; -ENOENT if key is unavailable but the filename isn't a
+ * correctly formed encoded ciphertext name, so a negative dentry should be
+ * created; or another -errno code.
  */
 static inline int fscrypt_prepare_lookup(struct inode *dir,
 					 struct dentry *dentry,
-					 unsigned int flags)
+					 struct fscrypt_name *fname)
 {
 	if (IS_ENCRYPTED(dir))
-		return __fscrypt_prepare_lookup(dir, dentry);
+		return __fscrypt_prepare_lookup(dir, dentry, fname);
+
+	memset(fname, 0, sizeof(*fname));
+	fname->usr_fname = &dentry->d_name;
+	fname->disk_name.name = (unsigned char *)dentry->d_name.name;
+	fname->disk_name.len = dentry->d_name.len;
 	return 0;
 }
 
diff --git a/include/linux/fscrypt_notsupp.h b/include/linux/fscrypt_notsupp.h
index 15bbbd7c52b8b..24b261e49dc14 100644
--- a/include/linux/fscrypt_notsupp.h
+++ b/include/linux/fscrypt_notsupp.h
@@ -112,7 +112,7 @@ static inline int fscrypt_setup_filename(struct inode *dir,
 	if (IS_ENCRYPTED(dir))
 		return -EOPNOTSUPP;
 
-	memset(fname, 0, sizeof(struct fscrypt_name));
+	memset(fname, 0, sizeof(*fname));
 	fname->usr_fname = iname;
 	fname->disk_name.name = (unsigned char *)iname->name;
 	fname->disk_name.len = iname->len;
@@ -199,7 +199,8 @@ static inline int __fscrypt_prepare_rename(struct inode *old_dir,
 }
 
 static inline int __fscrypt_prepare_lookup(struct inode *dir,
-					   struct dentry *dentry)
+					   struct dentry *dentry,
+					   struct fscrypt_name *fname)
 {
 	return -EOPNOTSUPP;
 }
diff --git a/include/linux/fscrypt_supp.h b/include/linux/fscrypt_supp.h
index b8443a2e47dc4..8641e20694ce9 100644
--- a/include/linux/fscrypt_supp.h
+++ b/include/linux/fscrypt_supp.h
@@ -191,7 +191,8 @@ extern int __fscrypt_prepare_rename(struct inode *old_dir,
 				    struct inode *new_dir,
 				    struct dentry *new_dentry,
 				    unsigned int flags);
-extern int __fscrypt_prepare_lookup(struct inode *dir, struct dentry *dentry);
+extern int __fscrypt_prepare_lookup(struct inode *dir, struct dentry *dentry,
+				    struct fscrypt_name *fname);
 extern int __fscrypt_prepare_symlink(struct inode *dir, unsigned int len,
 				     unsigned int max_len,
 				     struct fscrypt_str *disk_link);
-- 
2.29.1

Re: [PATCH 4.19 0/5] backport some more fscrypt fixes to 4.19

[Greg KH]

On Sat, Oct 31, 2020 at 03:05:48PM -0700, Eric Biggers wrote:
> Backport some fscrypt fixes from upstream 5.2 to 4.19-stable.
> 
> This is needed to get 'kvm-xfstests -c ext4,f2fs,ubifs -g encrypt' to
> fully pass on 4.19-stable.  Before, generic/397 and generic/429 failed
> on UBIFS due to missing "fscrypt: fix race where ->lookup() marks
> plaintext dentry as ciphertext".
> 
> This also fixes some bugs that aren't yet covered by the xfstests.
> E.g., "fs, fscrypt: clear DCACHE_ENCRYPTED_NAME when unaliasing
> directory" fixes a bug that caused real-world problems on Chrome OS.
> 
> Some relatively straightforward adjustments were needed to the patches,
> mainly due to the refactoring of fscrypt.h that was done in 5.1.

All now queued up, thanks!

greg k-h